The One, the True, the Good & the Beautiful: a philosophy of data visualisation.

Data visualisation, done well, shows “transcendental” attributes of existence such as existence, unity or particularity, truth, value, beauty, and prompts emotions of awe and even ecstasy.

Yes, really. Done well.

What do you mean, ‘graphs don’t make you feel like that’?

Let me see if I can convince you…

Before you begin

Before you visualise data, you should already have been through a process of exploration I call epistemology — starting with a survey of what exists, and moving through to understanding the nature and scope of the knowledge available. A data visulaistion epistemology should answer the questions: ‘what can we know?’… and just as importantly ‘who needs to know it?’ (the audience) and ‘when do they need to know it?’ (Kairos, the perfect moment).

Once you have surveyed the various data you’re already capturing; you can then go on to determine all of the decisions and actions (and inactions) these data could inform or prompt. It’s useful to distinguish between diagnostic and executive actions.

Diagnostic actions determine the component causes, and how much of the change each component explains, and whether each component can be considered a sufficient and/or necessary cause, and inferring the best explanation.

Executive actions are changes with hope of a resultant, measurable, effect.

Once you have chosen which of the actions you want to focus on in the visualisation, work backwards from the action to the data. Pull together a decently sized data sample, and you’re ready to get going on your visualisation.

The One…

So, what is it we’re looking at again? We need to be clear on how we are dividing up existence into definable, particular units — the fundamental unit of discourse. In a project management visualisation, for example, we would consider the task… or the team member. To visualise ecommerce performance we might start with the product… or alternatively the user goal. In an Educational context, perhaps the learning goal or the concept. The fundamental unit can be thought of as a boundary separating what the thing is from what it is not. You also need a way of uniquely defining and referring to each unit — a Taxonomy. You may also want to understand how each unit relates to other units — an Ontology. Finally, you also need to understand the space through which the unit moves, the fundamental dimension. This is almost always time, sometimes space, and very occasionally both.

The True…

Truth is not binary. There are lots of values in between TRUE and FALSE. Unfortunately, encoding data into a visualisation makes humans believe it more, so take care to expose uncertainty. In short, seeing is believing, so use this power wisely. Embrace your poor data quality. Learn the difference between absence, error, lie, and, worst of all, bullshit, and design your visualisation to make these differences in ‘truthiness’ apparent. Typical approaches include error and uncertainty bars or regions, as well as blurring, fading, colour coding and jitter. Ensure that you differentiate between what your data reports as hoped for (budgeted or planned) and what past data suggests is likely to happen (expected). Make clear the uncertainty inherent between the past and the future.

The Good…

What is the ‘goodness’, the intrinsic value of owning or controlling each unit? Often expressed as a monetary value, beware inflation when projecting forwards. Have an underlying time-specific value (e.g. 2016 pounds value) that can be used to scale to, even as the value of money changes over time. If you are measuring people, or anything countable, try to ensure that a given number of people has the same visual weight wherever it is expessed. This can give rise to challenges on geographic maps, for example, where the geographic area is not proportional to the number of people within it (so try a cartogram). You may also wish to think about different kinds of good — value to an organisation; value to all (the common good) — and different metrics you can use to get at the good: money, time, carbon-embodiment, gross national happiness. You may also wish to consider to apply your fundamental dimension… if this happens to be time, then can you show the difference between the good that is-now and the good that may-be, with the actions needed to get reach towards the shadow of future good?

The Beautiful…

If goodness (value) brings pleasure through possession, beauty gives pleasure simply in its beholding. If the visualisation is a pleasure to behold, it will encourage engagement and so understanding. Conversely, if it isn’t beautiful, it will not be used. So how to achieve beauty? Emphasise synthesis of information into a unified context, cognitive ease of understanding (relationships shown visually, logic presented verbally or in text), harmony of form and due proportion (height to width ratios of 1:1 — square and circle; 1:2; the harmonic 2:3 and 3:4 — cut one in half to get the other; and of course the famous self-harmonic golden ratio), colour, and the relationship of parts to whole.

… and finally, the Sublime

Beautiful things are easy to apprehend, that is, it makes it easy to lay hold of a thing mentally, so as to understand it clearly, at least in part. Ease of apprehending is pretty close to ease of comprehending: embracing or understanding it “in all its compass and extent”. Can the sheer scale of the knowledge available be communicated such that the viewer feels awe? Can we excite thoughts and emotions beyond ordinary experience? It should be like the first time you meet an Alp up close. This is not something that can be explained in text, you need to show it.

Epilogue: The Final Act

Finally, the quality test. Does the visualisation successfully prompt the actions we had defined at the beginning of the process?

Beauty should make the action clear. Goodness should make the action valuable. Truth should make the action confident. Touching the sublime should make the action… joyful.

Good luck.